Epistatic and gene-by-environment (GE) interactions are an important biological basis for most, if not all, complex diseases including drug addiction. The lack of powerful analytical methods has significantly limited our efforts to identify ubiquitous interactive contributors. In this R01 application, we propose: 1) to develop more effective combinatorial approaches for detecting multifactor interactions;2) to develop computer software for implementing the proposed methods;and 3) to apply the new methods to genetic studies for identification of important risk factors that predispose individuals to nicotine dependence (ND). We hope that by using a general class of statistics such as likelihood score (or score-like), generalized estimating equations score, principal component score, or other measures that reflect the association between the putative factors and the phenotype of interest, our new methods will allow justification of both discrete and continuous covariates. Furthermore, our methods are expected to be applicable to qualitative, quantitative, and multiple correlated traits and more flexible for use in various population-based and family-based study designs. The prospective software will be platform independent and easy to use and will run in both graphic-user interface and console forms. Eventually, our goal is to have methods and software that will offer a better option than the existing tools for detecting interactive and/or jointly active determinants. Finally, we propose to apply these newly developed methods to the existing 3072 SNP dataset of ND data in a sample of 602 nuclear families, consisting of 2037 subjects of either African-American or European-American origin to explore epistatic and gene and environment interactions systematically in ND. To our knowledge, such an application is the first attempt to investigate the role of interactions between genetic and non-genetic risk factors in smoking-related phenotypes more fully. Moreover, it will help to identify susceptibility genes from interactive biological pathways and environmental factors that are most relevant and to offer tailored preventive advice and therapeutic intervention to smokers. Given that tobacco smoking claims more than 435,000 US lives per year, identification of interactions in this project is highly relevant to the nation's public health agenda. PUBLIC HEALTH RELEVANCE: This project proposes to development novel methodology of detecting gene-gene and gene-environmental interaction, which represents an extension of our early work on the development of a novel, generalized multifactor dimensionality reduction method. Furthermore, we propose to implement our newly developed methods into a computer software package and use them to analyze an existing 3072 SNP dataset of nicotine dependence data in a sample of 602 nuclear families, consisting of 2037 subjects of either African-American or European-American origin.